The first AI-Observer Summer School was held at the Eratosthenes Center of Excellence in Limassol, Cyprus, from July 10-14. Training sessions were given by Prof. Fabio Del Frate, Giorgia Guerrisi and Lorenzo Giuliano Papale (Tor Vergata University of Rome), and Dr. Gerd Reis (German Research Center for Artificial Intelligence). During the five-day hybrid event, more than 50 participants learned about the application of artificial intelligence in Earth observation, with special focus on disaster risk management. Topics included deforestation, flood detection, and natural hazard management using Sentinel-1 Synthetic Aperture RADAR (SAR), and Sentinel-2 Multi-Spectral Imaging (MSI) data.

AI-OBSERVER consortium:
Eratosthenes Center of Excellence
Deutsches Forschungszentrum für Künstliche Intelligenz
Università degli Studi di Roma Tor Vergata
Cellock Ltd.

The Augmented Vision department and its partners are presenting three projects at the ICT 2015.
Come and visit:

Location: Off-site Area (Praça do Comércio, Lisbon)
Exhibition date: 18 – 22 October 2015
Project homepage: http://www.easy-imp.eu/

Location: Transform Area
Booth: T26
Exhibition date: 20 – 22 October 2015
Project homepage: http://eyesofthings.eu

Location: Innovate Area
Booth: i33
Exhibition date: 20 – 22 October 2015
Project Homepage: www.euromov.eu/alterego

A paper was accepted as oral presentation at the International Conference on Computer Vision (ICCV) 2015:

Christian Bailer, Bertram Taetz, Didier Stricker
Flow Fields: Dense Correspondence Fields for Highly Accurate Large Displacement Optical Flow Estimation

ICCV is one of the premier conferences for researchers in the field of computer vision.
The conference will be held from December 11th to September 18th in Santiago (Chile).

We are delighted to announce the winning of smartDoc competition (challenge 2) as part of the competitions in ICDAR 2015 Conference.

The goal of the competition is to extract the textual content from document images which are captured by mobile phones. The images are taken under varying conditions to provide a challenging input (full description of the challenge).

The method is based on “Combining Clustering and Classfication Results” (CCC); initially the background color is used to first detect and dewarp the document. The image is then binarized to extract lines, words and subwords. Those are then clustered incrementally across all the corpus. A 1D LSTM is trained on both sharp and blurry gray-scale text-lines for recognizing subwords. Clusters of subwords are labeled by majority voting. 

Each year an expert jury from the University of Kaiserslautern is nominating outstanding doctoral theses for several academic subjects together with the Kreissparkasse Kaiserslautern..

We congratulate Dr.-Ing. Petersen on winning this award for Computer Science.

This award was granted for his doctoral thesis that deals with the understanding of manual workflows from video examples. An important application of his research work is the automatic generation of interactive Augmented Reality manuals from the observation of a reference performance.

Links:

https://www.kskkl.de/module/ueber_uns/pressecenter/upload/preisverleihung_stiftung_fuer_die_tu_kl.pdf

Abstract  doctoral thesis:

Workflow knowledge comprises both explicit, verbalizable knowledge and implicit knowledge, which is acquired through practice. Learning a complex workflow therefore benefits from training with a permanent corrective. Augmented Reality manuals that display instructive step-by-step information directly into the user’s field of view provide an intuitive and provably effective learning environment. However, their creation process is rather work intensive and current technological approaches lead to insufficient interactivity with the user. In this thesis we present a comprehensive technical approach to algorithmically analyze manual workflows from video examples and use the acquired information to teach explicit and implicit workflow knowledge using Augmented Reality. The technical realization starts with unsupervised segmentation of single work steps and their categorization into a coarse taxonomy. Thereafter, we analyze the single steps for their modalities using a hand and finger tracking approach optimized for this particular application. Using explicit, work step specific generalization we are able to compensate for morphological differences among different users and thus to reduce the need for large amounts of training data. To render this information communicable, i.e., understandable by a different person, we present the further processed data using Augmented Reality as an interactive tutoring system. The resulting system allows for fully or semi-automatic creation of Augmented Reality (AR-) manuals from video examples as well as their context-driven presentation in AR. The method is able to extract and to teach procedural, implicit workflow knowledge from given video examples. In an extensive evaluation, we demonstrate the applicability of all proposed technical components for the given task.

Prof. Didier Stricker gave a key impulse talk about „Technologiy and Innovation“ in the area of Digital Technology at the MIT:Futura. The event analyzed barriers for digitalization, creation of start-ups and large capital investment in Germany. Investors, founders and European Commissar for Digitalization – Mr Oettinger – discussed current situation and possible changes for speeding-up technological changes at economical and societal level.

More on: www.ioxp.de

…is the newly started European project at AV.

The kick-off meeting was held on January 27th 2015 at the University of Castilla-La Mancha.

Read also the press release on the Kick-off meeting.

If you are interested in the project’s goals and objectives, partners and the idea behind it, please check also the project website http://eyesofthings.eu